TopoEEG: A Timesformer-Based Topographic Image Representation Method for Single-Trial Early Detection of P300

Detecting P300 event-related potentials (ERPs) from electroencephalography (EEG) finds applications in several domains including brain computer interfaces (BCIs). Due to the low signal-to-noise ratio (SNR), however, detecting P300 from single trials has been challenging. As a result, multiple repetitions are typically needed, extending the BCI decision time. To address this problem, in this paper, we present TopoEEG, a novel method that generates a sequence of topography-based images from EEG, to capture both spatial and temporal information. These sequences are subsequently processed using the state-of-the-art video classification technique, TimeSformer. Evaluated using the BCI Competition II (IIb) dataset, results demonstrate that TopoEEG achieves single-trial P300 detection with F1-scores of 0.6830 and 0.7134 within the first 500 ms and 600 ms, respectively. These results highlight TopoEEG's potential to enhance the efficiency and speed of BCIs by reducing the need for multiple repetitions and the time required for decision-making in single-trial P300 detection.